#!/usr/bin/env bash

# TODO: Add a comment block at the top of the script, describing what it does.
# TODO: Add a '-h' or '--help' flag, that displays information about the script, and how to use it.


#  sitegen / s4g
#  -------------
#
#  This is a static-site generator, that can be used to convert Markdown files into HTML.
#  It is extremely simple and extremely opinionated, as you can see if you read the code below.
#  A lot of paths and directories are hard-coded, to suit my workflow.
#
#  To use it, create a directory for your project (e.g. 'website'). Inside 'website', create
#  two directories: 'source' (which holds your Markdown files) and 'output' (which holds the
#  converted CSS.
#
#  In addition to these directories, three files are needed in 'website':
#  1. 'header.html' - A header, which is prepended to every source file. Unfortunately, this must
#		      be written in regular HTML.
#  2. 'footer.html' - A footer, which is appended to every source file. Also must be written in HTML.
#  3. 'styles.css' - A global stylesheet.
#
# If you have any comments or questions, please email aadhavan@twomorecents.org.


set -o errexit # Stop executing when a command fails
set -o nounset # Stop executing when accessing an unset variable
set -o pipefail # Treat a pipeline as failing, even if one command in the pipeline fails

if [[ "${TRACE-0}" == "1" ]]; then set -o xtrace; fi # Enable tracing (output of each command) if the TRACE variable is set



BASE_PATH="$(dirname "$0")"

check_for_dirs() {
	if [[ ! -d "${BASE_PATH}/source" ]]; then
		echo "ERROR: 'source' folder does not exist. Your content is sourced from this folder." >&2
		exit
	fi

	if [[ -d "${BASE_PATH}/temp" ]]; then
		echo "ERROR: You have an existing 'temp' folder. Please delete this folder, and run the script again." >&2
		exit
	fi

	if [[ ! -f "${BASE_PATH}/header.html" ]]; then
		echo "ERROR: You do not have a header.html file. This file is used as a global header. Please create this file, and run the script again." >&2
		exit
	fi
	if [[ ! -f "${BASE_PATH}/footer.html" ]]; then
		echo "ERROR: You do not have a footer.html file. This file is used as a global footer. Please create this file, and run the script again." >&2
		exit
	fi

}


setup_temp_dir() {
#	Check if 'temp' already exists
	mkdir "${BASE_PATH}/temp"
}

setup_output_dir() {
	rm -rf "${BASE_PATH}/output" # Delete existing 'output' directory
	cp -r "${BASE_PATH}/source" "${BASE_PATH}/output" #Copy directory structure from 'source' to 'output'
}

del_files_in_output() {
	find "$BASE_PATH/output" -type f -name "*.md" -delete #Delete all .md files (which were copied over from 'source') in 'output'
}

read_metadata() {
#	Read the metadata from the top of a .md file into a string
	metadata=$(awk 'BEGIN{RS = "\n\n"} {print $0}; {exit}' "$1")  # Reads from the .md file until a double-newline is encountered
}

convert_to_array() {
	local meta_key
	local meta_value

#	Converts the metadata into two arrays: one with the key, and the other with the value.
	readarray -t meta_key < <(echo -e "$1" | awk -F: '{print $1}')
	readarray -t meta_value < <(echo -e "$1" | awk -F: '{st = index($0,":"); values = substr($0,st+1); print values}' | cut -c 2-)

#	Merge both arrays into an associative array
	declare -Ag  meta_array
	for index in $(seq 0 `expr "${#meta_key[@]}" - 1`); do
		meta_array["${meta_key[$index]}"]="${meta_value[$index]}"
	done
}

add_date_to_array() {
	if [[ "${meta_array[date]-}" == "auto" ]]; then # If the date is set to 'auto'
		meta_array["date"]="$(date -r $1 +'%b %d, %Y')"
	fi
}

add_header_and_footer() {
#	Copy header to temporary location
	cp "$BASE_PATH/header.html" "$BASE_PATH/temp/temp_header.html"

#	Check for relevant metadata, and perform corresponding action
	if [[ ! -v "meta_array["date"]" ]]; then # If there is no date
		sed -i '$ d' "$BASE_PATH/temp/temp_header.html" # remove the 'date published' section of the header
	fi

	if [[ "${meta_array[noappend]-}" == "true" ]]; then
		sed -i 's/ - Two More Cents//g' "$BASE_PATH/temp/temp_header.html" # 'noappend' removes the suffix from the title
	fi

#	Add header
	cat "$BASE_PATH/temp/temp_header.html" | cat - "$1" > "$BASE_PATH/temp/temp.html"

#	Add footer
	echo >> "$BASE_PATH/temp/temp.html" # Add newline
	cat "$BASE_PATH/footer.html" >> "$BASE_PATH/temp/temp.html"

#	Move temp file to original location
	mv "$BASE_PATH/temp/temp.html" "$1"
}

add_header_and_footer_to_index() {
#	Add header
	cat "$BASE_PATH/header.html" | head -n -1 | cat - "$1" > "$BASE_PATH/temp/temp.html" # For the index page, remove the last line of the header (date published)

#	Add footer
 	echo >> "$BASE_PATH/temp/temp.html" # Add newline
	cat "$BASE_PATH/footer.html" >> "$BASE_PATH/temp/temp.html"

#	Move temp file to original location
	mv "$BASE_PATH/temp/temp.html" "$1"
}

replace_vars() {
#	Loop through the keys of the 'meta_array' array, search for all occurences of the key in the HTML doc, and replace them with the corresponding value..
	for arr_key in "${!meta_array[@]}"; do
		meta_array["$arr_key"]="${meta_array["$arr_key"]/\//\\/}" # Escape all forward slashes in the value
		sed -i "s/[\$][\$]$arr_key[\$][\$]/${meta_array[$arr_key]}/g" "$1"
	done

}

md_to_html() {
#	Convert .md files from 'source' and place them into the correct locations into 'output'
#	Since the conversion is the slowest part of the script, I run each conversion process
#	in a separate thread (first loop). Then I wait for all of them to finish, and add the 
#	header and footers and replace the variables in each converted document (second loop).
	local files=$(find "$BASE_PATH/source" -name "*.md")

	for file in $files; do

		echo "Converting $file"

		read_metadata "$file" # Sets the 'metadata' variable

#		Copy file to temp dir and strip metadata
		cp "$file" "$BASE_PATH/temp/"
		let num_lines=$(echo "$metadata" | wc -l)+1
		sed -i "1,${num_lines}d" "$BASE_PATH/temp/$(basename "$file")"

#		Construct path for output file
		local path_for_output=$(realpath --relative-to="${BASE_PATH}/source" "$file")
		path_for_output="${BASE_PATH}/output/${path_for_output}"
		path_for_output="$(dirname $path_for_output)/$(basename $path_for_output .md).html"

#		Convert the file (using the given filters), and place the output in the correct location.
#		The process is forked, so that all documents can be converted concurrently.
		(pandoc --lua-filter "$BASE_PATH"/pandoc_filters/* -f markdown --wrap=preserve "$BASE_PATH/temp/$(basename "$file")" > "${path_for_output}") &

		unset metadata
	done

#	Wait for all documents to finish converting, then remove all temporary files.
	wait
	rm "$BASE_PATH"/temp/*
	
	for file in $files; do
#		This function was already called above in the previous loop.
#		However, I need to call it again, since I need the metadata 
#		again for each file to create the 'meta_array' variable.
		read_metadata "$file" # Sets the 'metadata' variable

		convert_to_array "$metadata" #Sets the 'meta_array' array
		add_date_to_array "$file" #Uses 'meta_array' array

#		Construct path for output file. This is the same path that the converted
#		document is written to, in the loop above.
		local path_for_output=$(realpath --relative-to="${BASE_PATH}/source" "$file")
		path_for_output="${BASE_PATH}/output/${path_for_output}"
		path_for_output="$(dirname $path_for_output)/$(basename $path_for_output .md).html"

		add_header_and_footer "$path_for_output" # Uses 'meta_array' array
		replace_vars "$path_for_output" #Uses 'meta_array' array

		unset metadata meta_key meta_value meta_array
	done
}


gen_sorted_file_list() { # Generate a list of the HTMl files, sorted by when they were last modified (read from the contents of the HTML file)
	local files=$(find "$BASE_PATH/output" -name "*.html")
	local date_mod

	for file in $files; do
		if grep -q "date-published" "$file" ; then
			echo "$file" >> "$BASE_PATH/temp/file_listing.txt" # Write files that have a date published to a temp file (we only want the files with date modified, because only these files can be listed with their date on the site map)

			date_mod+=$(cat "$file" | grep "date-published" | awk -F'[<>]' '{print $3}' \
				| cut -d' ' -f '1,2' --complement | tr -d "," | awk '{print $2" "$1" "$3}' \
				| date -f - +"%s")
#			Explanation:
#			Line 1 extracts the published date from the HTML file
#			Line 2 re-arranges this information, and converts it into DD MM YY format
#			Line 3 converts this into a UNIX timestamp

			date_mod+=$'\n'
		fi
	done

	date_mod=$(echo "${date_mod-}" | head -n -1) # Remove last (empty) line from variable
	echo "${date_mod-}" > "$BASE_PATH/temp/date_mod.txt" # Write the corresponding 'date modified' timestamps to a temp file

	paste "$BASE_PATH/temp/file_listing.txt" "$BASE_PATH/temp/date_mod.txt" > "$BASE_PATH/temp/new_file_list.txt" # Combine file list and date modified into a single file

	sorted_file_list=$(sort -r -k 2 "$BASE_PATH/temp/new_file_list.txt") # Sort the data in the file based on the timestamp (from newest to oldest), and store it into a variable
	sorted_file_list=$(echo "$sorted_file_list" | awk '{print $1}') # Store only the first column (the file path) in the variable
}

gen_index_page() { # Generate an index page (site map) that includes links to the other pages

	echo "Generating index page..."

	local index_file_html="<nav class=\"toc\">"$'\n' # Variable to store the body HTML of the index page; enclose the list in a nav

	for file in $1; do
		local title=$(cat "$file" | grep "<title>" | head -n 1 | awk -F'[<>]' '{print $3}') # Find the title of the web page
		local suffix=" - Two More Cents"
		title="${title%"$suffix"}" # Remove the website name from it

		local pub_date=$(cat "$file" | grep "date-published" | head -n 1 | awk -F'[<>]' '{print $3}') # Find the date published
		prefix="Published on " # Find date published of webpage
		pub_date="${pub_date#"$prefix"}" # Remove the prefix from it
		pub_date=$(echo "$pub_date" | tr -d "," | awk '{print $2" "$1" "$3}' | date -f - +"%m/%d/%Y") # Re-arrange the date and convert to mm/dd/yy

		local file_path=$(realpath --relative-to="${BASE_PATH}/output" "$file")

		index_file_html+="<li><time>${pub_date}</time> - <a href=\"$file_path\">$title</a></li>" # Add a line of HTML containing the date and title of the article
		index_file_html+=$'\n'
	done

	index_file_html=$(echo "$index_file_html" | head -n -1) # Remove last (empty) line from variable
	index_file_html+="</nav>"

	path_for_output="${BASE_PATH}/output/site-map.html"
	echo "$index_file_html" > "$path_for_output" # Output variable to file

	add_header_and_footer_to_index "$path_for_output" # Add header and footer to index file
	sed -i 's/[\$][\$]title[\$][\$]/Site Map/g' "$path_for_output" # Replace title variable with 'site map' title

}

copy_things_in() {
	cp "${BASE_PATH}/styles.css" "${BASE_PATH}/output/"
	cp -r "${BASE_PATH}/files" "${BASE_PATH}/output/"
	cp -r "${BASE_PATH}/fonts" "${BASE_PATH}/output/"

}

clean_up() {
	rm -r "${BASE_PATH}/temp"
}


check_for_dirs
setup_temp_dir
setup_output_dir
del_files_in_output
md_to_html
gen_sorted_file_list # Sets the 'sorted_file_list' variable
gen_index_page "$sorted_file_list" # Uses the 'sorted_file_list' variable
copy_things_in
clean_up